An implementation of the Epsilon-Greedy Algorithm for use with BLAND
To take advantage of this code, it is necessary to follow the steps listed at https://github.com/scattering/pycrysfml/blob/master/doc/Ubuntu_deps.txt within a linux terminal.
The files inside this repository are intended to be located and run within the pycrysfml package.
Specifically, all of these files should be placed within a folder of pathname: pycrysfml/hklgen/examples/sxtal/foldername, where one substitutes their foldername of choice for the word 'foldername'.
Provided one already has a .cfl and .int file premade, one should place those files into the folder a single tier above the code in this repository.
For example, all of our code was inside folder pycrysfml/hklgen/examples/sxtal/epGreedy, while the .int and .cfl files were inside pycrysfml/hklgen/examples/sxtal.
There are a number of changes to be made to file epGreedy.py:
- line __: Substitute the names of your .int and .cfl files into the pathnames for variables 'observedFile' and 'infoFile'.
observedFile = os.path.join(DATAPATH,r"../<your filename here>.int")
infoFile = os.path.join(DATAPATH,r"../<your filename here>.cfl")- line __: Alter the parameters to suit your variable of interest.
General formatting involves:
m.atomListModel.atomModels[index of your atom of interest].[parameter of interest (coordinate)].value = initial guess
m.atomListModel.atomModels[index of your atom of interest].[parameter of interest (coordinate)].range(lowerBoundaryOfUnitCell, upperBoundaryOfUnitCell)
For example, for our simulated MoTe2 trials, these lines were:
m.atomListModel.atomModels[0].z.value = 0.5
m.atomListModel.atomModels[0].z.range(0,1)